Innate Molecular and Cellular Signature in the Skin Preceding Long-Lasting T Cell Responses after Electroporated DNA Vaccination

Vaccination is the most attractive strategy to prevent and control infectious diseases (1). However, there are still no effective vaccines against many infectious diseases, including emerging and chronic infections. Modern vaccine strategies have focused on the development of recombinant vectors (2). DNA-based vaccines are an attractive approach (3) to induce immune responses (4–6) by mimicking viral Ag production, including endogenous processing and presentation (7). The main limitation for the use of DNA vaccines in humans has been their low immunogenicity. To overcome this issue, various approaches have been developed in animal models (4, 6, 8–12, 13) and in humans (14–16). Some of these approaches consist in improving the plasmid vector design to enhance induced immune responses (17–19). Krohn et al. (20), for example, developed an auxoGTU DNA vector that encodes the E2 bovine papillomavirus protein in addition to vaccine Ags. The E2 protein facilitates persistence of the plasmid in host cells during cell division (20, 21), leading to a more efficient Ag production over time (20) and a stronger immune response stimulation compared with a conventional CMV plasmid vector (22). Other approaches focused on improving the vaccine delivery into the cells to enhance DNA vaccine efficiency. Electroporation (EP) has notably been developed. EP facilitates the entrance of the DNA vaccine into host cells through short electric impulsions delivered at the vaccinated site that temporarily destabilize the cell membranes (13, 23). This method has dramatically enhanced the immunogenicity of DNA vaccines in nonhuman primates (NHPs) (24–26) and also in humans. Indeed, two therapeutic DNA vaccines delivered by the i.m. route with EP have shown promising results in the context of cervical intraepithelial neoplasia in two-thirds of patients, with up to complete regression of lesions in some patients (27, 28).

Another important parameter in the immune response induction is the site of vaccine administration. Indeed, several studies highlight the influence of route of immunization in the orientation of immune responses (29, 30). The skin appears to be an attractive organ for vaccination because of the presence, already at steady state, of various subtypes of innate immune cells, including APCs (31). Indeed, the high density of APCs in the skin is probably related to the lower vaccine doses required for immune response induction relative to other vaccination routes, providing a cost advantage for vaccines delivered intradermally (32, 33). Langerhans cells (LCs), the unique steady-state resident APC population of the epidermis (34, 35), can be mobilized under specific inflammatory conditions (36) and are involved in vaccine responses (37). The dermis is inhabited by several dendritic cell (DC) subsets that have a distinct capacity to produce cytokines, cross-present Ags, and induce T cell proliferation and differentiation (38–41). Resident macrophages, the main APC subset of the dermis (42–44), have a key role in maintaining homeostasis (45) and actively participate in the protection against infection by phagocytizing pathogens and secreting soluble factors (46, 47).

Despite the intense use of vaccination, the immune mechanisms that lead to protection are still poorly understood (48). Furthermore, an increasing number of studies suggest that early events are critical for vaccine efficiency (49–51). We previously demonstrated that intradermal auxoGTUmultiHIV DNA vaccination in combination with EP enhances polyfunctional CD4 and CD8 T cell responses in NHPs (25). We showed that EP affects the localization and intensity of vaccine Ag production in the skin (25, 52) and enhances LC mobility, facilitating their interaction with vaccine Ag-expressing cells and their departure from the epidermis (52). In this study, we investigated the early cellular and molecular events that occur after intradermal administration of the auxoGTUmultiSIV vaccine combined with EP. We show that this vaccine induced local inflammation, with a recruitment of an inflammatory cell population at the vaccinated site. We characterized the soluble factors produced in the local microenvironment and the expression of genes involved in the innate immune response. We highlighted that in these processes EP and the DNA vaccine stimulate different components of the innate immunity that may act in symbioses to induce the strong and persistent T cell responses characteristic of this vaccination strategy.

Twenty-three adult male cynomolgus macaques (Macaca fascicularis) were imported from Mauritius to study early-state (nine animals) and adaptive (14 animals) responses. The macaques weighed between 4 and 9 kg. Animals were housed in the Commissariat à l'Energie Atomique et aux Energies Alternatives (CEA) facilities (accreditation number B 92-032_02), and handled in accordance with European guidelines for NHP care (European Union Directive N 2010/63/EU). The CEA complies with the Standards for Humane Care and Use of Laboratory Animals of the Office for Laboratory Animal Welfare under Office for Laboratory Animal Welfare Assurance number A5826-01. This study was approved by the regional committee for the use and care of animals (Comité Régional d’Ethique Ile de France Sud, reference 11_013). Before the beginning of the study, the animals were confirmed to be seronegative for several pathogens (SIV, simian T lymphotropic virus, filovirus, hepatitis B virus, herpes B, and measles). Animals were sedated with ketamine (10–20 mg/kg, Imalgene 1000; Rhone-Mérieux, Lyon, France) and 10% of acepromazine (Vetranquil, Ceva Santé Animale, Libourne, France) during handling.

The DNA auxoGTUmultiSIV vaccine was obtained from FIT Biotec (Tampere, Finland). The vaccine was diluted in PBS (Life Technologies, Paisley, U.K.) to a final concentration of 1 mg/ml. After being shaved, the macaques were immunized with the auxoGTUmultiSIV plasmid by intradermal injection of 100 μl vaccine formulation or PBS via a 29-gauge needle (Myjector 1ml; Terumo, Leuven, Belgium). Each animal received the DNA vaccine at one site of the skin and the PBS at another site to permit paired comparisons of subsequent events; the vaccines were immediately followed with an EP performed as previously described (53) Skin biopsies (8 mm in diameter) were performed on anesthetized animals 1, 3, and 8 d after injection.

Cells were extracted from fresh skin specimens using modified versions of published protocols (54, 55). Briefly, the s.c. fat tissue was removed, and the specimens were then incubated with 4 mg/ml bacterial protease Dispase II (Roche Diagnostic, Meylan, France) in PBS with 1% 100× of penicillin–streptomycin–neomycin (Life Technologies) and 0.25 μg/ml of Fungizone (amphotericin B; Life Technologies) for 12–16 h at 4°C and then for 40 min at 37°C. Epidermal and dermal sheets were separated and incubated in 2 mg/ml Collagenase D (Roche Diagnostics) and 0.2 mg/ml DNase I from bovine pancreas (Sigma-Aldrich) in RPMI 1640 at 37°C with shaking for 20 and 40 min, respectively. Next, the epidermis was incubated with 0.25× of trypsin (Eurobio, Courtaboeuf, France) for 10 min, the remaining tissue was mechanically dissociated with tweezers, and the cellular suspension obtained was filtrated (100 μm). The dermal tissue was further mechanically dissociated with a gentleMACS Dissociator (Miltenyi Biotec), and the obtained cellular suspension was filtrated (100 μm).

Cell mortality was assessed using the LIVE/DEAD Fixable Dead Cell Stain Kit (Life Technologies) according to the supplier’s instructions. Nonspecific Ab staining was blocked by incubation with a 5% solution of pooled macaque sera. The following mAbs were used for the characterization of immune cells: HLA-DR (clone L243; Becton Dickinson [BD]), CD1a (clone O10; Dako, Glostrup, Denmark), CD3 (clone SP34-2; BD), CD45 (clone DO58-1283; BD), CD8 (clone RPA-T8; BD), CD20 (clone L27; BD), CD209 (clone DCN46; BD), CD14 (clone M5E2; BD), CD83 (clone HB15e; BD), CD86 (clone FUN-1; BD), CD163 (clone GHI/61; BD), CD11b (clone Bear 1; Beckman Coulter), CD123 (clone 7G3; BD), CD206 (clone 19.2; BD). CD66abce (clone TET2; Miltenyi Biotec, Bergisch Gladbach, Germany), CD1c (clone AD5-8E7; Miltenyi Biotec), and CD207 (clone 2G3; Baylor Institute for Immunology Research, Dallas, TX). Unlabeled Abs were detected with a secondary Ab coupled to an Alexa Fluor fluorochrome with the Zenon Antibody Labeling Kit (Life Technologies). For the detection of intracellular proteins, cells were incubated in Cytofix/Cytoperm solution (BD) before the staining with Ab diluted in Perm/Wash buffer (BD). Acquisition was performed on a BD LSRFortessa cytometer (BD), and the obtained data were analyzed using FlowJo 9.7.1 software (Tree Star, Ashland, OR).

The percentage of skin cells is represented by the mean ± SD. Data were analyzed using Prism 5.0 (GraphPad, La Jolla, CA). The Friedman test with Dunn posttest was used to assess the significance of differences in cell frequency over time. The Wilcoxon test was used to compare differences in the cell frequency between vaccinated sites and controls.

For cytokine analysis, the media of epidermal and dermal cells were obtained by collecting the supernatant during the cell extraction phase and stored at −80°C. Cytokine concentrations were measured using the MILLIPLEX MAP Non-Human Primate Immunoassay Kit (Millipore, Guyancourt, France), according to the supplier’s instructions. The concentration (in picogram per milliliter) of all 22 tested soluble factors is shown in Supplemental Tables I and II. An ANOVA test with the Bonferroni posttest was used to compare cytokine concentrations between vaccine- and PBS-injected sites, both with EP.

ELISpots were performed using MultiScreen 96-well filtration plates (Millipore) prepared as previously described (25). Briefly, the plates were coated by incubation overnight with 10 μg/ml mAb against monkey IFN-γ (clone GZ-4; Mabtech AB, Sophia Antipolis, France) in PBS at 4°C. Plates were washed with PBS, blocked with culture medium supplemented with 10% heat-inactivated FCS (culture medium; Laboratoires Eurobio), and 2 × 10⁵ PBMCs were added to each well. Fifteen-mer overlapping peptides (11 aa overlap), tailored according to the sequence of the MultiSIV protein, were then added in triplicate to a final concentration of 2 μg/ml for each peptide in the culture medium. The peptides were subdivided into four different subpools covering the sequence of Rev (24 peptides), Nef (63 peptides), Tat (29 peptides), and Gag p15/27 (85 peptides). Phorbol 12-myristate 13-acetate (Sigma-Aldrich) and ionomycin (Sigma-Aldrich) at final concentrations of 0.1 and 1 μM, respectively, were used as a positive control. Culture medium alone was used as a negative control. Plates were incubated for 18 h at 37°C in a humid atmosphere containing 5% CO₂ and washed before an overnight incubation at 4°C with 1 μg/ml biotinylated anti–IFN-γ Ab (clone 7-B6-1; Mabtech AB). Plates were washed again and incubated with 0.25 μg/ml alkaline phosphatase–streptavidin conjugate (Sigma-Aldrich) for 1 h at 37°C before a final washing step. Spots were developed by adding 80 μl of NBT/BCIP substrate (Sigma-Aldrich) to each well and counted with an automated ELISpot reader system with KS software (Carl Zeiss, Le Pecq, France). The results are expressed as the mean of IFN-γ spot-forming cells (SFC) per 1 × 10⁶ PBMCs (IFN-γ SFC/million PBMCs) of triplicate wells. The background was calculated as the mean number of IFN-γ SFC/million PBMCs in nonstimulated samples. Samples yielding more than 50 IFN-γ SFC/million PBMCs after removal of the background were scored as positive.

Whole-skin RNA was extracted from macaque skin biopsy specimens and stored in RNAlater using TissueRuptor followed by the RNeasy Plus Universal Kit (QIAGEN), according to the manufacturer’s instructions. The quality of the total RNA was verified using an Agilent 2100 Bioanalyzer. RNA quantity was measured using a NanoDrop ND-1000 spectrophotometer. Cyanine-3–labeled cRNA was prepared from 200 ng of total RNA using the Quick Amp Labeling Kit (Agilent Technologies) according to the manufacturer’s instructions, followed by RNeasy column purification (QIAGEN, Valencia, CA). Dye incorporation and cRNA yield were measured using a NanoDrop ND-1000 spectrophotometer. Cyanine-3–labeled cRNA (1.65 μg) was fragmented at 60°C for 30 min in a reaction volume of 55 μl containing 1× Agilent fragmentation buffer and 2× Agilent blocking agent, following the manufacturer’s instructions. Upon completion of the fragmentation reaction, 55 μl of 2× Agilent Hybridization Buffer was added to the fragmentation mixture, and the mixture was hybridized to Agilent Rhesus Macaque Gene Expression Microarrays v2 for 17 h at 65°C in a rotating Agilent Hybridization Oven. After hybridization, microarrays were washed for 1 min at room temperature with GE Wash Buffer 1 (Agilent) and 1 min at 37°C with GE Wash Buffer 2 (Agilent). Slides were scanned immediately after washing on an Agilent DNA Microarray Scanner (G2505C) using a one-color scan setting for 4 × 44K array slides (scan area, 61 × 21.6 mm, scan resolution 5 μm, dye channel set to green, the photomultiplier tubes voltage set to 100%). The scanned images were analyzed with Feature Extraction software 10.7.3.1 (Agilent) using default parameters.

Transcriptomic signals were background corrected using the Robust Multi-array Average method and quantile-normalized. Differentially expressed genes were identified using a t test (p value <0.01) and a fold-change threshold of 1.5. Functional enrichment analyses of biological functions and upstream regulators were performed using Ingenuity Pathways Analysis software (Ingenuity Systems). Ingenuity Pathways Analysis maps each gene identifier to its corresponding molecule in the Ingenuity Pathways Knowledge Base. The p values generated by the Fisher exact test were adjusted using Benjamini–Hochberg multiple testing for all analyses. The multidimensional scaling (MDS) representation was generated using the singular value decomposition–MDS algorithm (56). MDS methods aim to represent the similarities and differences among high-dimensionality objects in a space with a low number of dimensions, generally 2 or 3, for visualization purposes (57). Pairwise distances between the dots are proportional to the Euclidean distances between the samples. Biological conditions are indicated by convex hulls (i.e., the smallest convex set containing the points). The Kruskal Stress criterion (57), shown in the representation, quantifies the quality of the representation as a fraction of the information lost during the dimensionality reduction procedure.

Raw transcriptomic data used in this study are available on the European Bioinformatics Institute ArrayExpress database (http://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-8973. Microarray data are also available on the Infectious Disease Models and Innovative Therapies (IDMIT) data dissemination platform (http://data.idmitcenter.fr/DNAvaccination-EP/).

In a previous work, we described the immunogenicity of the auxoGTUmultiHIV vaccine. We showed that the specific design of this DNA vaccine associated with EP induces a strong and persistent specific recall cellular responses against the HIV Ag encoding by the plasmid (25). In this study, we aimed to characterize the innate events at the vaccine injection site to understand the link between these events with the adaptive immune response induction in the context of the auxoGTUmultiSIV vaccine. We first compared the immunogenicity of the auxoGTUmultiSIV vaccine delivered with or without EP in two groups of animals injected intradermally at weeks 0, 4, and 12 (Fig. 1). We measured specific T cell responses through the release of IFN-γ by ELISpot in vaccinated animals.

As expected, the EP group showed an enhanced adaptive response relative to the non-EP group. The strongest responses were directed against Nef, followed by Gag, Tat, and Rev. Cumulative measurements showed that the magnitude of the adaptive response was very high after the third immunization, with a persistent vaccine response that lasted at least 30 wk after the first immunization.

To study the effect on skin cells of auxoGTU DNA vaccine associated with EP, we first characterized immune cell subsets present in macaque epidermis and dermis at baseline. Epidermal and dermal sheets were enzymatically separated, and each skin layer was processed for cytometry analysis. Dead cells, debris, and CD45⁻ cells were excluded from the analysis to characterize innate immune cell populations (Supplemental Fig. 1).

LCs were identified in the epidermis by their high expression of CD1a and HLA-DR. We confirmed our previous finding (34), showing that, in contrast to human LCs, macaque LCs are negative for CD1c (40, 58) (Supplemental Fig. 1A).

Few CD66-expressing granulocytes were detected in the dermis at baseline (Supplemental Fig. 1B). Furthermore, we detected dermal CD1a⁺CD163^low cells, which could be subdivided into two subsets. The main CD1a⁺CD1c⁺ subset corresponded to dermal DCs, according to previous publications (34, 36). The minor dermal CD1a⁺ subset expressed high levels of CD1a but not CD1c, a phenotype that closely resembles LCs, suggesting that they are LCs migrating through the dermis. Dermal macrophages were identified based on their expression of CD14 and their high expression of both HLA-DR and CD163.

These analyses identified different APC populations present at baseline in macaque skin. To characterize the cell abundance and phenotype changes occurring over time after auxoGTU DNA vaccination in combination with EP, we then performed a kinetic and extended phenotypical analysis.

To assess cell recruitment after vaccination, skin biopsy specimens were collected at baseline day 0 (d0), d1, d3, and d8 after PBS injection combined with EP (PBS/EP) and DNA injection combined with EP (DNA/EP), respectively (Fig. 2).

In the epidermis, PBS/EP and DNA/EP induced a significant influx of polymorphonuclear leukocytes (PMNs) at d1 (p = 0.0008 and p < 0.0001, respectively), which declined at d3 (p = 0.0061 and p = 0.0487, respectively), and a significant influx of CD14⁺HLA-DR⁺ cells at d1 (p = 0.0011 and p = 0.0008, respectively) and d3 after PBS/EP (p = 0.0084) (Fig. 2A).

In the dermis, both PBS/EP and DNA/EP treatment induced an influx of PMNs at d1 (p = 0.0031 and p = 0.0061, respectively) with a significantly stronger influx at the DNA/EP site compared with the PBS/EP site (p = 0.0391). The frequency of PMNs reduced by d3 but was still significantly increased in the PBS/EP group (p = 0.0370). We did not observe significant changes compared with the baseline among the CD14⁺HLA-DR⁺ subset in the dermis at any time point (Fig. 2B).

These data suggest that EP is the main inducer of the local inflammation following vaccination, with only a slight influence of the DNA vaccine. To confirm this hypothesis, we measured the frequencies of PMNs and CD14⁺HLA-DR⁺ cells in the epidermis and dermis under four conditions: PBS, PBS/EP, DNA, DNA/EP at d1 and d3 (Supplemental Fig. 2). In the epidermis, even if NS, probably because of the low number of animals per group (n = 3), we confirmed a similar recruitment of PMNs at the PBS/EP and DNA/EP sites, whereas a very low increase of this population occurred at the PBS and DNA sites. In the dermis, a stronger recruitment of PMNs at the EP site occurred compared with the non-EP sites. Even if less clear, the same observation could be made for the CD14⁺HLA-DR⁺ subset.

Altogether, these results suggest that EP plays a major role in the early induction of local inflammation, whereas the auxoGTU DNA vaccine, per se, only influences this process slightly.

We observed an influx of CD14⁺HLA-DR⁺ cells following vaccination in the epidermis but not in dermis. We decided to further characterize the phenotypes of these cells at baseline and following injection and EP (Fig. 3). CD14⁺HLA-DR⁺ cells were completely absent from the epidermis and formed a homogeneous population in the dermis at baseline, whereas additional CD14⁺HLA-DR⁺ phenotypic subsets, based on CD11b and CD163 expression, appeared after PBS/EP and DNA/EP in each of the skin sheets.

At baseline, resident macrophages, defined as CD163^highCD11b⁺ (subset 1), were present only in the dermis. After injection and EP, a second subset, expressing a higher level of CD11b and a lower level of CD163 compared with resident dermal macrophages, and a third subset of CD11b⁺CD163⁻ cells arose (Fig. 3A). The second and third subsets are likely newly recruited cells because of their total absence in the skin at baseline and their presence in the epidermal and dermal compartments after both PBS/EP and DNA/EP. Indeed, we measured a significant influx of the CD11b^highCD163^mid subset (subset 2) at d1 (p = 0.0031) and d3 (p = 0.0031) post–PBS/EP as well as post–DNA/EP but with significance only at d1 (p < 0.0001) (Fig. 3B). There was also a significant recruitment of the CD11b⁺CD163⁻ subset (subset 3) that peaked at d1 (p = 0.0014) post–DNA/EP. However, no changes in the resident dermal macrophage population (subset 1) could be noticed after injection and EP. Because resident macrophages represent the main CD14⁺HLA-DR⁺ population, even after injection and EP, it may explain why only a nonsignificant trend of increase of this population was measured when considering the globality of the CD14⁺HLA-DR⁺ subsets (Fig. 2B).

To further characterize these three CD14⁺HLA-DR⁺ subsets, we performed additional surface staining by focusing on the d1 post–DNA/EP condition in the dermis (Fig. 3C). The resident dermal macrophages (subset 1) were CD206⁺CD40⁺CD86^highCD11b⁺HLA-DR⁺. This phenotype is similar to a type 2 macrophage (M2) (47, 59). The second subset was instead CD206⁻CD40⁻CD86⁺HLA-DR⁺, a phenotype closer to a type 1 macrophage (M1) (60). The third subset also lacked expression of CD206 and showed the lowest expressions of HLA-DR and CD86, suggesting that these cells were the least activated. Indeed, this third subset resembles circulating monocytes (Supplemental Fig. 3B), suggesting that these cells might be monocytes circulating in the dermis or in a process of differentiation.

Altogether, these results show that EP, per se, induced a rapid recruitment of two additional CD14⁺HLA-DR⁺ populations to the vaccinated skin that were distinct from resident macrophages.

We previously showed that EP enhances LC mobility in the epidermis by in vivo microscopy, facilitating the interaction of LCs with Ag-transfected cells (52). In this study, we further investigated the effect of the auxoGTUmultiSIV DNA vaccine combined with EP regarding DC subpopulations in the skin.

At baseline, in the epidermis, only the CD1a-expressing LCs could be observed, whereas two CD1a-expressing subsets were present after PBS/EP and DNA/EP: 1) the first subset expressing the highest level of CD1a and negative for CD1c corresponds to LCs; and 2) the second subset expressed CD1a at an intermediate level and was positive for CD1c (Fig. 4A). We studied the dynamics of these two CD1a-expressing cell types over time after vaccination (Fig. 4B). We observed a transient increase of LC frequency at d1, with significance only at the DNA/EP site (p = 0.0016). At the same time point, LC strongly upregulated their expression of CD86, CD83, and HLA-DR compared with the baseline, illustrating their activation (Fig. 4C). This transient increase was followed by a significant decrease in the LC frequency between d1 and d3 at both the PBS/EP (p = 0.0061) and DNA/EP (p = 0.0370) sites, suggesting migration of LCs out of the epidermis. A comparison of the PBS/EP and DNA/EP conditions at d1 and d3 revealed no significant changes in LC frequency due to the presence of the auxoGTU DNA vaccine (Fig. 4B).

The CD1a^intCD1c⁺ subset appeared in the epidermis at the PBS/EP and DNA/EP sites at d1 (DNA/EP p = 0.0221) and d3. A comparison of the PBS/EP and DNA/EP conditions showed a significant effect of the auxoGTU DNA vaccine regarding the recruitment of this subset at d1 (p = 0.0288) and d3 (p = 0.0011) (Fig. 4B). No staining of CD207 was observed at the surface of the CD1a^intCD1c⁺ cells, demonstrating that these cells were distinct from LCs. Nevertheless, ∼20% of these cells showed CD207 expression in the intracellular compartment, suggesting they may share some LC characteristics (Fig. 4D).

In the dermis, we did not noticed any effect of the PBS/EP or DNA/EP treatment on DC populations (Supplemental Fig. 4).

These results demonstrate that EP induced LC activation as defined by increase expression of HLA-DR, CD86, and CD83 at d1. Subsequently, the activated LC might migrate out of the epidermis, illustrated by an LC frequency decrease between d1 and d3. The auxoGTU vaccine seems to have no to little impact in this process but enhances the recruitment of a second CD1a-expressing cell subset that, even if distinct from LCs, may share some LC characteristics.

Our results suggest that EP is the main inducer of the local cell recruitment, with only a slight influence of the auxoGTU DNA vaccine. We now decided to study the local effect of the vaccine on the molecular microenvironment. We analyzed soluble factors released by epidermal and dermal cells after PBS/EP or DNA/EP treatment using multiplex assays (Fig. 5).

We observed the release of proinflammatory soluble factors by both epidermal and dermal cells, such as MCP-1, IL-8, IL-13, and IL-15, at each time point postinjection ex vivo (Fig. 5, Supplemental Tables I, II). Interestingly, we observed a clearly enhanced release of MCP-1 at d1 and IL-15 at d1, d3, and d8, both in the epidermis and dermis, and MIP-1β, IL-18, and TNF-α, primarily in the dermis, in the presence of the DNA vaccine. Anti-inflammatory soluble factors, such as IL-10, IL1-RA, and sCD40L, were also upregulated, mainly in the dermis and only in the presence of the DNA vaccine. For example, sCD40L levels significantly increased (p = 0.0306) at d8 at the DNA/EP injection site, whereas there was no change of this factor at the PBS/EP injection site. Several other soluble factors, such as IL-1β remained under the detection threshold of the device, both in epidermis and dermis over the kinetics (Supplemental Tables I, II).

These results highlighted the effect mediated by the auxoGTU DNA vaccine on the initiation of the proinflammatory and concomitantly balanced anti-inflammatory soluble molecular response induced in the skin during the early phase with this vaccination strategy.

To further characterize the molecular events involved in the auxoGTU DNA vaccination strategy combined with EP, we performed a transcriptomic analysis of skin biopsy specimens collected at d1, d3, and d8 after PBS/EP and DNA/EP and at an untreated site (baseline) using microarrays (Fig. 6).

Among the 43,603 genes on the microarrays, 3963 (9.09%) were differentially expressed in at least one condition relative to baseline. Consistent with the observed early cellular events, the strongest transcriptomic responses were observed at d1 for both PBS/EP and DNA/EP conditions, with 2655 (6.09%) and 1131 (2.59%) genes differentially expressed, respectively (Fig. 6A). Fig. 6B shows an MDS representation of the transcriptomic profiles calculated based on the list of 3963 genes that were differentially expressed in at least one condition. Each dot in the MDS represents a biological sample, and the distances between dots are proportional to the transcriptomic distances between the samples. Consistent with the numbers of differentially expressed genes found at each time point, we observed that the PBS/EP and DNA/EP samples from d1 postinjection were displaying the most differential responses relative to baseline. The PBS/EP and DNA/EP conditions overlapped in the MDS representation at d3 and d8, suggesting similar responses. In contrast, the profiles for the PBS/EP and DNA/EP conditions were nonoverlapping at d1, suggesting that different molecular events were involved at the earliest timepoint.

We identified two sets of 1549 and 1826 genes. The first set was upregulated throughout the time course, whereas the second set was downregulated. Canonical pathway analysis of these gene clusters revealed that upregulated genes were significantly associated with cyclins and cell cycle regulation, p38 MAPK signaling, and TREM1 signaling (Supplemental Fig. 4A). These pathways are involved in cell division, inflammation, and neutrophil and monocyte/macrophage activation, respectively. We also noticed that canonical pathways involved in pattern recognition, acute phase response signaling, and TLR signaling were also upregulated. These pathways are involved in Ag or danger signal recognition. Genes involved in IFN signaling were also strongly upregulated, especially at d1 at the DNA/EP site.

The analysis of upstream regulators (Supplemental Fig. 4B) showed the upregulation of CSF2, involved in granulocyte and monocyte/macrophage production, differentiation, and function; ERBB2, a member of the epidermal growth factor receptor family of receptor tyrosine kinases; and TNF, involved in various inflammatory responses. Several genes encoding transcription factors, such as FOXM1, MYC, TP63, and the NFKB complex, were upregulated, whereas other transcription factor genes, such as TP53, NUPR1, and KDM5B, were downregulated after both PBS/EP and DNA/EP.

The transcriptomic analysis showed that both PBS/EP and DNA/EP vaccination strategies induced the strongest changes in gene expression at d1 postinjection. The changes in gene expression induced by the PBS/EP and DNA/EP injections were associated with cellular division, influx, and activation, as well as immune responses. The strongest differences between the PBS/EP and DNA/EP transcriptomic profiles occurred at d1.

To focus on genes involved in the innate immune response, we next restricted the transcriptome analysis to cytokines/chemokines and their receptors and activation markers differentially expressed relative to baseline (Fig. 7). We also focused on the transcripts differentially expressed between the DNA/EP and PBS/EP conditions.

Several genes involved in cell migration, such as CCL3, CXCL10, CXCL11, CCL1, CCL5, CCL11, and IL8, were the most significantly upregulated relative to baseline at both the PBS/EP and DNA/EP sites, especially at d1 and d3 (Fig. 7A). IL20, a cytokine produced by keratinocytes during skin inflammation, was also upregulated at d1 and d3 at both the PBS/EP and DNA/EP injection sites. Several genes involved in APC activation, such as CD80 and CD86, were significantly upregulated at d1 and d3, consistent with the phenotypic analyses (Fig. 4C). Interestingly we noticed a stronger upregulation of several of these genes in the presence of the DNA vaccine. For example, IRF7, a gene encoding the IRF7 transcription factor, involved in virus-inducible cellular genes, was more strongly upregulated at d1 post–DNA/EP than after PBS/EP.

To highlight the specific contribution of the auxoGTU DNA vaccine in the gene perturbation after injection and EP, we performed a comparison of the gene expression profiles between the PBS/EP and DNA/EP at each time point. Only 184 genes were differentially expressed between the PBS/EP and DNA/EP conditions (Fig. 7B). Among the upregulated genes in DNA/EP, we found several IFN-inducible genes, such as IFIT3, IFIT5, IRF7, ISG15 or MX1. Furthermore, we found an upregulation of serum amyloid A (SAA4) and CCL3, CCL5, CXCL11, and CXCL10, suggesting that the auxoGTU DNA vaccine significantly affected the expression of several specific genes involved in the immune response and, notably, in the IFNs signaling pathway.

Finally, we assessed which pattern recognition receptors (PRRs) were involved in this DNA vaccine strategy by studying the gene expression of TLRs 1–10 and adaptor molecules (Fig. 8A). Surprisingly, we observed an upregulation of the TLR2 and the TLR3 at d1 post–DNA/EP, whereas no change occurred in the expression of TLR9 able to recognize unmethylated CpG motifs contained in plasmid DNA. The contribution of TLR9 in DNA vaccine sensing is a matter of debate, and some studies have shown that TLR9 is dispensable in vaccine plasmid sensing, suggesting the contribution of other DNA sensors. We examined the transcription of cytosolic DNA sensors and observed the highly significant upregulation of AIM-2 at d1 only at the DNA/EP site (Fig. 8B). Upon binding to dsDNA, AIM-2 forms the inflammasome, leading to the cleavage of pro–IL-1β and -18 by caspase-1 and the secretion of their active forms. AIM-2 is involved in the DNA-sensing inflammasome, but other inflammasome structures can be formed involving, notably, NOD-like receptor as NLRP3 and NLRP4. We did not observe enhanced expression of NLRP3 or NLRP4, but we highlighted a significant increase of CASP1, the gene coding the caspase-1 only at d1 post–DNA/EP (Fig. 8C). These data suggest a specific activation of the DNA-sensing inflammasome involving AIM-2.

It has been recently demonstrated that AIM-2 is an essential sensor for the immunogenicity of DNA vaccines in mice (61). Indeed we observed a trend of positive correlation between the relative AIM-2 expression and several IFN-inducible gene expression at d1 post–DNA/EP (Fig. 8D). Even if not significant, probably because of a fewer number of individuals in this analysis, these trends suggested a link between the auxoGTU DNA vaccine sensing via AIM-2 and the enhancement of immune system activation.

Overall, these results demonstrate that the DNA/EP vaccination strategy affected cell division, cell migration, and cellular activation. In addition, even if the EP is able to induce gene expression change, the auxoGTU DNA vaccine is required to strongly enhance these changes. Indeed, the DNA vaccine specifically enhanced the expression of AIM-2, a DNA sensor involved in inflammasome formation and subsequent innate inflammatory responses.

Innate immunity has been shown to play a key role in the orientation of adaptive immune responses (50, 51, 62), suggesting its impact on vaccine efficiency (48). We have previously shown that auxoGTUmultiHIV DNA vaccination delivered with EP induces strong specific T cell responses that remain detectable for at least 2 years in animal blood (25). In this study, we characterized the innate responses induced locally following vaccination with the auxoGTUmultiSIV vaccine delivered with EP (DNA/EP) and at a PBS/EP site as control to understand the mechanisms involved in the subsequent T cell responses.

We first characterized the local inflammation by flow cytometry and showed an early influx of CD66⁺ granulocytes and HLA-DR⁺CD14⁺ monocytes/macrophages at d1 and d3 in the epidermis and in the dermis at both the PBS/EP and DNA/EP sites. A deeper phenotypic analysis of the HLA-DR⁺CD14⁺ subset showed that this population is heterogeneous after vaccination. Indeed, two additional populations of HLA-DR⁺CD14⁺ cells, absent at baseline, appeared after PBS/EP and DNA/EP: the CD163^midCD11b^high and CD163⁻CD11b⁺ cells. These two subsets are distinct from the CD163^highCD11b⁺–resident dermal macrophages, whose frequency remains stable following PBS/EP and DNA/EP. The CD11b^highCD163⁺ subset resembles proinflammatory M1 macrophages based on their lack of CD206, a marker associated with tissue and M2 macrophages, and their recruitment under inflammatory conditions (59, 60, 63). However, they also express some markers usually associated with M2 macrophages, such as a high level of CD11b and a lower level of CD40 and CD86 compared with resident macrophages. These results confirmed the high plasticity of macrophages (60, 63–65), in which macrophages share markers associated with the M1 or M2 type, depending on the inflammatory condition. The CD163⁻ cells may correspond to blood monocytes recently recruited in the skin following injection and EP (66, 67). These CD163⁻ cells could be in a process of differentiation into inflammatory macrophages (CD163⁺CD11b^high) or inflammatory DCs or may serve as blood precursors to repopulate the skin of resident populations as demonstrated in different studies (44, 68, 69). Nevertheless, further investigations are required to confirm the accurate contribution of these monocytes/macrophages during the early phase of the auxoGTU with EP vaccination strategy and determine the ontological relationship between these populations.

In parallel, we observed in the epidermis the influx of an additional population of CD1a-expressing cells, distinct from CD1a^highHLA-DR^high LCs. These cells expressed CD1a at a lower level than LCs and, in contrast to LCs, expressed CD1c and did not express CD207 at their surface. They may correspond to inflammatory dendritic epidermal cells, which were described in the human epidermis (70, 71) and dermis (72, 73) upon inflammation. However, a proportion of these CD1a^int cells expressed CD207 in the intracellular compartment, indicating that they may also be related to LCs. Indeed, their recruitment in an inflammatory context suggests that they might even be LC precursors, as described in several mouse studies (74, 75). In addition, these CD1a^int CD1c⁺ cells increased CD1a and decreased CD1c expression over time after vaccination to acquire a phenotype closer to LCs, strengthening the hypothesis that these cells are LC precursors.

We previously showed that EP influences the intensity and localization of Ag expression in the skin (25). Without EP, Ag expression was exclusively produced in the dermis, whereas with EP, Ag expression was still found in the dermis but also in large amounts in the epidermis, especially in the superior differentiated keratinocyte layers (52). At baseline, LCs are localized close to the basement membrane. Previously, we have shown by in vivo fiber confocal microscopy (76) that DNA vaccination with EP increased LC mobility and induced their migration through the superior epidermal layer, where Ag was produced in the highest amounts, whereas immunization without EP did not affect LCs localization or density (25). In this study, we confirmed the effect of EP on LC mobility through a transient increase of LCs in the epidermis at d1 at DNA/EP and PBS/EP sites. This increase was associated with the upregulation of CD86, CD83, and HLA-DR, illustrating LC maturation, and followed by a decrease in LC frequency by d3, suggesting a migration of LCs to the skin-draining lymph nodes (77).

Otherwise, this study highlighted that similar events occurred at the PBS/EP and DNA/EP injection sites. These observations strongly suggest that EP was the main inducer of these events, with only a marginal contribution of the auxoGTU DNA vaccine. Initially, EP was used to enhance DNA entrance into cells to increase Ag production and improve subsequent specific immune responses (78–80). However, our data together with the results of other groups (78, 80–83) suggest that the role of EP is not limited to an increase of Ag synthesis. Actually, EP causes low-intensity tissue damage, which rapidly resolves, leading to inflammatory response. This inflammatory response was suspected to be essential for the activation of the immune system and enhancement of the adaptive response, and that, independently of the Ag quantity increase due to EP (80, 81). Peng et al. (80) demonstrated in a mouse hepatitis B virus DNA vaccine model that EP applied before the DNA vaccine injection resulted in an enhanced Ab response similar to, or even stronger than, the response obtained when EP was applied just after DNA administration. Additionally, our transcriptomic data showed that several genes involved in skin cell proliferation and tissue repair were differentially expressed after both PBS/EP and DNA/EP. FOXM1, for example, is a transcription factor known to play a key role in cell cycle progression. This protein is required for the proper execution of mitosis and has an important role in tissue repair after injury in adults (84, 85). We also observed an upregulation of TP63 and SPDEF. TP63 has a crucial role in skin development and maintenance (86) and is strongly upregulated in hyperproliferative epidermal cells during normal skin wound healing in mice. SPDEF is mainly associated with cancer, but Yang et al. (87) reported a significant upregulation of SPDEF in expanded skin relative to normal skin in humans. Altogether, these upregulated genes may thus have a role in skin repair, notably by inducing cell proliferation in response to the low tissue damage induced by EP.

Even if the cytometry analysis did not reveal a strong difference between PBS/EP and DNA/EP, a deep analysis of soluble proteins and transcript expressions highlighted subtle modifications on the innate response. Indeed, the analysis confirmed at the protein level that different responses were induced at the PBS/EP and DNA/EP injection sites. The DNA vaccine specifically stimulated skin cells by enhancing their ability to release soluble factors. MCP-1, IL-15, IL-18, and sCD40L were produced in significantly higher amounts in at least one skin compartment and mostly in the dermis. A trend of increased production of IL-8, MIP-1β, IL-10, IL-1RA, and TNF-α also occurred in the presence of the DNA vaccine. Chemokines and cytokines are essential for inducing cell recruitment and their activation as well as the orientation of the immune response (88). Notably, DC maturation is crucial for T cell activation. In the absence of proper inflammatory signals, DCs can migrate and present Ag to T cells without being fully mature. Under these conditions, DCs can induce T cell anergy or direct differentiation to tolerogenic T cells (89, 90). IL-15 is crucial for CD8⁺ T cell activation (91–93). In this study, we found that IL-15 was produced only in the presence of the auxoGTU DNA vaccine. We hypothesize that the activation of LCs by EP, in combination with enhanced IL-15 release due to the auxoGTU vaccine, may be crucial for the strong activation of CD8⁺ T cells. Overall, these results suggest that EP and the auxoGTU DNA vaccine differentially stimulate skin cells and act in concert to activate innate immunity (78).

The analysis of the transcription profile confirmed our multiplex results. Indeed, several genes encoding chemokines, such as CXCL10, CXCL11, CCL5, and IL8, were induced with a stronger intensity at the DNA/EP site. Additionally, we show at the transcriptomic level that a large number of genes involved in immune responses (CCL3, SAA-4, or SIRT-6) and different IFN-inducible genes (STAT2, MX1, IFIH1, and OAS2) that play a crucial role in antiviral responses were significantly upregulated in the presence of the auxoGTU DNA vaccine.

The auxoGTU vectors are expressed in the Escherichia coli expression system and therefore contain CpG motifs. DNA vaccines mimic viral infection, and our results demonstrate that this vaccination strategy induces a specific transcriptional antiviral signature, probably in response to PRR stimulation by the auxoGTUmultiSIV vaccine, leading to IFN production. However, the characterization of the PRRs involved in this DNA vaccine recognition needs to be further investigated. Our microarray data suggest a TLR involvement, as we observed a stronger upregulation of MyD88 at the DNA/EP than at the PBS/EP site at d1. However, there was no change in TLR9 expression, which plays a role in the recognition of unmethylated CpG motifs contained in bacterial plasmids, suggesting that this TLR is not involved in this vaccination strategy. Although several studies have shown that TLR9 was required for DNA vaccine efficacy (94, 95), others have demonstrated that TLR9 was dispensable for DNA vaccine recognition (96–98) and specific immune response induction, suggesting other receptors were involved in DNA vaccine sensing. We also observed an upregulation of TLR3 at d1 postvaccination albeit only in presence of DNA. TLR3 is a dsRNA sensor localized in the endosome. RNA polymerase III has also been reported to recognize foreign dsDNA and convert it into dsRNA, which can trigger RIG-I located in the cytoplasm (99). Our transcriptomic data showed an upregulation of the gene POL3RK encoding for the RNA polymerase III both at the PBS/EP and DNA/EP sites. We can hypothesis that the auxoGTU DNA vaccine was recognized and converted into dsRNA by the RNA polymerase III, which can trigger dsRNA sensors such as TLR3. These results together with the specific antiviral transcriptome signature might suggest an indirect adjuvant effect of the auxoGTU DNA vaccine through the stimulation of the TLR3 axis.

Interestingly, in this study, among cytosolic DNA sensors, AIM-2 was the only gene strongly upregulated at d1 in the presence of DNA. This gene encodes a PRR directly involved in dsDNA recognition. After binding to dsDNA, AIM-2 shapes one form of the inflammasome, leading to the cleavage of pro–IL-1β and -18 to their active forms by caspase-1. Interestingly, our results showed an upregulation of CASP1, the gene coding the caspase-1 but only in the presence of the auxoGTU vaccine. AIM-2 is an important DNA sensor involved in the efficacy of DNA vaccination strategies, and its depletion dramatically decreased humoral and cellular Ag-specific responses in an influenza DNA vaccination mouse model (61). It was also recently demonstrated that stimulating the AIM-2 axis enhances long-lasting specific CD8⁺ T cell responses (100). In this study, we also observed trends of positive correlations between the transcription of AIM-2 and several IFN-inducible genes such as IFIT3, MX1, MX2, HERC5, or OAS2. These data tend to confirm, in an NHP model, the involvement of AIM-2 in the sensing of DNA vaccine and the activation of the immune system by enhancing the antiviral response axis through IFN-inducible genes.

In conclusion, we demonstrated that this vaccination strategy allowed an efficient stimulation of innate immunity in which EP and the auxoGTU DNA vaccine differentially stimulated local cells, leading to a combined activation of innate immunity. EP acted as a strong adjuvant able to induce inflammatory cell recruitment and LC mobilization, whereas the auxoGTUmultiSIV vaccine seemed to be required to provide a specific signal to skin cells, probably through various PRRs (potential direct involvement of AIM-2 and indirect involvement of TLR3), leading to the release of soluble factors, such as IL-15, and the activation of antiviral immunity through the upregulation of several IFN-inducible genes. Altogether, these events may be required for a proper DC activation, resulting in the strong stimulation of the adaptive cellular immunity.

This work benefited from the technical support of the Animal Science and Welfare, the Laboratory of Immunology and Infection, and the Laboratory FlowCyTech core facilities of the IDMIT infrastructure. We thank Susann Fält and David Brodin at the Bionformatics and Expression Analysis core facility at Karolinska Institutet for technical support. We thank all members of the IDMIT infrastructure for excellent expertise and outstanding contributions.

The authors have no financial conflicts of interest.